Asbestos Fiber Type In Malignant Mesotheiioma : An Analytical Scanning Electron Microscopic Study of 94 Cases

Transcription

1 American Journal of Industrial Medidne 23 : (1993) 10 + < JfQI sturil s, 1st and), 966): I. In c and /./ / Asbestos Fiber Type In Malignant Mesotheiioma : An Analytical Scanning Electron Microscopic Study of 94 Cases Victor L. Roggli, MD, Phllip C. Pratt, fdo, and Arnold R. Brody, PhD h der tiolen and.ima- 7afety 991) : 219- lation t Anal actor aring Although the association between asbestos exposure and malignant rnesothelioma is indisputable, controversy continues regarding the relative eonltibption of the various types of asbestos fibers to the development of Inesothelioma. We examined the types of asbestos fibers recovered from lung parenchyrna in more than 90 cases of malignant mesothelioma from the United States, using an analytical scanningelectron microscope. Almost half of the patients were former asbestos insulators os.shipyard workers. The fibers were recovered from lung tissues obtained at autop ;y or surgical resection by means of a sodium hypochlorite digestion procedure. Atnoslte asbestos was identified in 81% of the cases and accounted for 58% of all fibers 5 µm or greater in length. Trernolitelactinolitelanthophyltite were identified in 55% of the cases and accounted for 10% of all fiber types. Chrysotile was identified in 21%of the cases and accounted for 3% of fibers exceeding 5 µm in length. Ctiocidolite was found in 16% of the eases and accounted for 3% of fibers exceeding 5 µm in length. Nonasbestos mineral fibers (commonly found in the lungs of the general population) were observed in 71% of the cases and accounted for 25% of all fibers 5 p.m or greater in length. The findings in this study are at odds with the assertion that crocidolite asbestos is responsible for most mesotheliomas in the United States. - O 1993 waey-l3ss. Inc. Key words: lung lissue burden, amosite, eroddolilq thrysotile, tremolite auma rte lo INTRODUCTION The association between exposure to asbestos and subsequent development of malignant mesothelioma is indisputable. However, controversy continues regarding the types of fibers that contribute to the development of inesothelioma. Conventional wisdom has taught us that all of the commercially valuable fotms of asbestosincluding amosite, crocidolite, and chrysotile along with its contaminant, tremoliteare capable of producing mesothetiomas in humans and experimental animals, but that the potency of the various fiber types differs with regard to their potential for producing mesotheliomas [Hammar and Bolen, 1988 ; Hillerdal, 1983 ; Roggli et al., This so-called fiber gradient concept suggests that crocidolite is the most Durham Veterans Administration and Duke University Medical Centers, Durham. NC (V.L.R.. P.C.P.). National Institute of Environmental Health Saiences, Research Triangle Park, NC (A.R.B.). Address reprint requests to Victor L. Roggli, MD, Dept. of Pathology, Box 3712, Duke University Medical Center. Durham, NC Accepted for publication July 10, Viicy-Liss, Inc.

2 606 Roggli et at. potent, chrysotile the least potent, and amosite somewhere in between with regard to their ability to produce mesotheliomas in humans, although no such gradient has been observed in experimental animal studies [Wagner et al., 1974). Nonetheless, not all investigators have agreed with this conventional wisdom, with some taking the position that there is no evidence for differences in fiber potency with respect to trw sothelioma [Nicholson et al., 19901, whereas others have argued that pure chrysotile does not produce mesotheliomas at all in humans [Craighead and Mossman, 1982 ; Mossman et al., 1990). Some have taken the extreme position that crocidolite asbestos alone is the cause of practically all tnesotheliomas In the world, including most cases in the United States [Wagner, 1991a,b). The authors have had the opportunity to examine the mineral fiber content of the lungs in more than 90 patients with malignant mesothelioma from the United States. These cases were referred from throughout the country, and thus represent a wide variety of exposures. However, a large proportion of the cases were either insulators or shipyard workers [Roggli, 1991], which have accounted for a substantial percentage of mesotheliomas seen in this country. The purpose of the present study is to examine the relative proportion of the various mineral fiber types that were present in the lung either at the time of thoracotomy for diagnosis or at auiopsy (usually within a year of diagnosis) in this group of patients with malignantsmesothelioma. MATERIALS AND METHODS The authors' files were reviewed for cases of malignant mesothelioma for which lung tissue was available for analysis of mineral fiber content by means of analytical scanning electron microscopy (SEM). The diagnosis of malignant mesothelioma was confirmed by one of the authors (VLR) in all but three cases using our previously published criteria [Roggli et al., 19871, which include consideration of the gross distribution of tumor, histologic appearance, and histochemical and immunohistochemical findings. Histochemical studies using periodic acid Schiff (PAS) with and without diastase predigestion and alcian blue with and without hyaluronidase predigestion and/or immunohistochemical studies using antibodies directed against cytokeratins, carcinoernbryonic antigen (CEA), or Leu-M I were performed in most cases and ultrastructural studies were performed in a few cases [McCaughey et al., 1991). In the three cases noted above, no histologic slides or blocks of tumor were referred to the authors' laboratory, but the diagnosis of mesothelioma was confirmed at autopsy at another institution. Records were reviewed for information regarding age, sex, location of tumor (pleural or peritoneal), and exposure history. The diagnosis of mesothelioma was made independent of asbestos exposure history or tissue mineral fiber content in all cases. Analysis of tissue mineral fiber content was performed using a JSM-35C SEM equipped with a Kevex energy dispersive spectrometer using techniques described previously [Roggli, 1989b ; Roggli et al., Lung tissue samples that had previously been fixed in formalin were digested in 5.25% sodium hypochlorite solution and the residue collected on 0.4 µm pore-size Nuclepore filters. The filters were mounted on carbon discs with colloidal graphite and sputter-coated with gold prior to examination by SEM. From 5 to 50 fibers were analyzed by means of energy dispersive x-ray analysis (EDXA) in each case ; in most cases, fibers were examined (only f ve fibers were analyzed in a few cases because of very low fiber

3 i,o-,l- i j 1 een all porlfr -tile 182; xstost VICC CROCIDDCI1E B RR S 60SEC lin u 2W H IBIEu 1 10 aa 10tEU IC Mesothelioma and Asbestos Fiber Type 607 the tts. ride tots :ntsto :t in thin I :CI e ee l l~~ f ' 1e 24, 1 Fig. 1. A. Energy dispersive X-ray specwm for amasite shows a prominent peak for siliedl, an intemediate peak for iron, and smalkr peaks for magnesium and, in this ease,'manganne. B. Spccuum for crocidoliie shows a prominent peak fm sil ;con, an intemxdiale peakifor iron, and smaller peaks for sodium and magnesium. iich ical W 1. roa, stoand edi- yto- Ises ) 1]. trul J at ige, s of eral EM :bed had soiters ;old :rgy vere iber counts and difficulty finding fibers 5 µm or greater in length ; 50 fibers were analyzed in two cases because of a special study). Fibers were identified and examined consecutively using a screen magnification of 1,500 x. Quantitative analyses for asbestos body content and uncoated fiber concentration for fibers S µm or greater in length were also performed in each instance, and these data have been reported previously for most of the cases [Roggli, 1991 ; Roggli et al., 1986, 1992). They.are not considered further in the present study. Fibers were defined as mineral particles with a length-to-diameter (aspect) ratio of at least three-to-one and roughly parallel sides. Only fibers that were greater than or equal to 5 µm in length were included in the analysis. At the magnification used, most of the fibers observed were 0.2 µm or greater in diameter. Fibers were classified as amosite, crocidolite, chrysotile, tremolite, actinolite, anthophyllite, or nonasbestos mineral fibers based on their morphology and elemental composition as determined from energy dispersive x-ray spectra [Roggli, 1989b, 1991 ; Roggli et al., 1986]. Stahdard spectra were obtained for comparison using the U.I.C.C. (International Union Against Cancer) asbestos standard samples obtained from Dr. V- Timbrell [Roggli et al., These standards include samples of amosite, crocidolite, chrysotile, and anthophyllite asbestos. The spectra for amosite and crocidolite are somewhat similar, being distinguished by a prominent magnesium peak for amosite and a prominent sodium peak for crocidolite (Fig. I). For fibers recovered from the lungs of mesothelioma patients, spectra were collected on each fiber until sufficient numbers of magnesium or sodium X-rays had been accumulated to distinguish between amosite and crocidolite. In a few instances, the distinction could not be made due to the presence of a substantial chlorine peak along with the sodium. Under this circumstance, it is not possible to determine whether the sodium peak derives from crocidolite or from contamination of the Glter with residual sodium hypochlorite

4 608 Roggli et at. TABLE 1. Demographtc and Palhotogte Information for 94 Patients With Malignant Mesothelioma No. AGE (yr) Median 60 Range SEX Mak 84 Female 10 LOCATION Pleunt 89 Peritoneal 5 SOURCE Autopsy 75 Surgical resection 19 EXPOSURE CA7E(IORY' Insulator 24,. Shipyard worker (other than insulator) 22 Other axbeslos 23.r Household contacts S ~ Building occupants 3 Other 5 Unknown 12 'See text for details. adherent to organic residues [Roggli, 1989b]. Such fibers are classified as amositel crocidolite (Amos/Croc) to acknowledge this confounding factor in f_iber identification. RESULTS The demographic data for our 94 cases of malignant mesothelioma are summarized in Table I and are similar to those reported in other large series of patients with mesothelioma [Alberts et al., 1988 ; Chahinian et al., 1982 ; Ruffle at al., 1989]. The median age for the 94 cases is 60 years (range years) ; 84 of the patients were men. In 89 cases, the tumor was of pleural origin, as opposed to a peritoneal origin in only five instances. Lung tissue for asbestos analysis was obtained at autopsy in 75 cases and from surgical resections in 19. A history of asbestos exposure was elicited in 74 or the 94 eases (79%). Twenty-four of the patients were asbestos insulators, a category that includes the job descriptions of insulator, pipecoverer, pipefitter, boiler worker, asbestos sawer, and asbestos sprayer. Twenty-two patients were shipyard workers other than insulators, including such job descriptions as joiner, welder, rigger, engineer, estimator, sandblaster, machinist, carpenter, mason, fireman, service in the U.S. Navy or merchant marine, and shipyard worker not otherwise specified. Twenty-three patients had histories of exposure to asbestos other than as insulators or shipyard workers, and these included asbestos cement workers, chemical maintenance workers, asbestos textile workers, sheet metal workers, asbestos factory workers, electricians, construcr I

5 Mesothelioma and Asbestos Fiber Type 609 TABLE It. Number and Percentage or Casea In Whieh Various Fiber Types Were Identified In 94 Patients With Malignant Mesothetioma Fiber type No. eeses Percent Amosite 76 BI Tremolitelamhophyllite/actinolite Chrysotile Crocidoiite Amos.R:roc' 5 5 Other 'Cases in which amosite could not be distinguished fnxn eroeidoau with certainry by means of EDXA. "Nonasbestos mineral fibers including talc, mtik, silica, aluminum silicates, other silicates, iron oxides, aluminum oxides, iron<hromium, fibrous glass. :mn :rel :r X sumpatients 1989]. patients ritoneal tined at (79%). the job cer, and ulators, r, sanderchant nts had Ts, and.sbestos mstruction workers, chemical engineers, nuclear or electrical power plant workers, jewelers, carpenters, painters and spacklers, inspection engineers,.railroad workers (during steam locomotive era), brake repair workers, and industrial ptposure to asbestos not otherwise specified [Roggli. 1991). Five additional patie4fs'were household contacts of asbestos workers, and these accounted for half of the cases that occurred in women. For three patients, the only identified possible source of exposure was working or attending school in a building with asbestos-containing materials. Five patients had no identifiable exposure to asbestog, including a truck driver, a district sales manager, and a teacher. Information regarding exposure to asbestos was unavailable in the remaining 12 cases. The number and percentage of the 94 cases among which each of the various fiber types were identified are summarized in Table 11. Amosite was the most prev alent fiber type, identified in 76 cases (81 %). The noncommercial amphiboles, tremolite, anthophyllite, or actinolite were identified in 52 cases (55%). The vast majority of these fibers had the typical composition of Si-Mg-Ca, which is the chemical signature for tremolite. Chrysotile fibers were detected in 20 cases (21 %). Crocidolite was identified in only 15 of the 94 cases (16%). Among these 15 cases, there was only one in which crocidolite accounted for a greater percentage of the fibers than either amosite, chrysotile, or the noncommercial amphiboles. This case was an 82-year-old man who was a railroad worker and who had insulated his own home in the 1940s. There were five cases (5%) in which commercial amphiboles were identified, but a reliable distinction between amosite and crocidolite could not be made. Nonasbestos mineral fibers, including talc, rutile, silica, aluminum silicates, other silicates, iron oxides, aluminum oxides, iron chromium, or fibrous glass (in order of decreasing frequency), were identified in 67 cases (71%). A total of 1,512 fibers were analyzed and identified by means of EDXA in these. 94 cases, an average of 16 fibers per case. More than half (58%) of the total fibers identified had the characteristic chemical signature for amosite asbestos (Fig. IA). An additional 10% of the fibers were noncommercial amphiboles, mostly tremolite. Only 3% of the fibers were crocidolite, which was similar to the percentage of chrysotile fibers detected. For 2% of the fibers analyzed, a reliable distinction between amosite and crocidolite could not be made. Nonasbcstos mineral fibers accounted for 25% of

6 Roggli et al. TABLE lll. Number and Percentage of Mineral Fiber Tyl>rs Among 1,512 Fibers Identified by EDXA Fiber type No. fibers Peramage Amosite 879 S8 Tmmolitdantlrophyllite/actinolite Crocidolite 43 3 Chrysaile 40 3 Amos./Croc.' Other 'See footnotes to Tabk 11. the total. Most of these were relatively low aspect ratio fibers, with lengths less than 10 µm and diameters of I µm or greater. DISCUSSION The present study indicates that among patients with Inalignant mesothelioma from the United States with a wide variation of exposure'fiistories, amosite is by far the most predominant fiber type recovered from the lung for fibers 5 µm or greater in length. Crocidolite, by comparison, was identified about 1/20 as often (Table III). It is widely believed that fibers in this size r}nge are the most relevant for the development of mesothelioma (Lippman, 1988 ; McDonald et al., 1989; Stanton et al., 1981 ; Wagner and Pooley, 1986). Previous studies from our laboratory had presented data on fiber analysis as commercial amphiboles, with amosite and crocidolite combined [Roggli, 1989b, 1991 ; Roggli et al., 1986, 1992]. These findings are comparable to those of other investigators from the United States and abroad. Churg and Wiggs [1984] reported analyses from 10 patients with malignant mesothelioma from the Pacific Northwest who had amphibole-induced malignancies. In nine cases, the ratio of amosite to crocidolite ranged from 2.3 to 37 (median value of 14.3). In only one case did the concentration of crocidolite fibers exceed that of amosite. This patient was an asbestos insulator from Warnock (1989] analyzed fiber burdens from 27 U.S. shipyard and construction workers with mesothelioma. The median amosite concentration was 1.2 million per gram of dry lung, whereas the median crocidolite concentration was million per gram (a ratio of about 40 :1) for fibers 0.25 µm or greater in length. In only two of the 27 cases did the concentration of crocidolite fibers exceed that of amosite. Mc- Donald et al. (1989] reported results of lung tissue analyses from 78 Canadian patients with mesothelioma. As compared to a reference population, (he risk increment and attributable risk for long (> 8 µm) amphibole fibers were 93.7 and 28% for amosite as compared to 24.9 and 10% for crocidolite in their group of patients with mesothelioma. Gaudichet ct al. [1988] analyzed lung tissue samples from 20 French subjects with malignant mesothelioma and found a significant difference between amphibole fiber retention in mesothelioma cases vs. controls, but no preponderance of crocidolite over amosite. Rogers et al. [1991] analyzed fiber burdens from 221 Australian cases of mesothelioma and found the greatest risk to be associated with the presence of long (? 10 µm) crocidolite fibers, which these authors attributed to the

7 Mesothelioma and Asbestos Fiber Type 611 ian ma far iter 11). the t et,iad w te~ 4th ced 37 lers 80. ion per per t of %Iclian :refor +ith nch cen nce?2i the the wide distribution of Wittenoom crocidolite throughout Australia. However, a significant association between risk and amosite fiber burdens was also reported. Noncommercial amphiboles were the next most commonly identified asbestos fiber type in our study, accounting for 10% of all fibers (Table III). Most of these were tremolite fibers, with smaller amounts of actinolite and anthophyllite. Tremolite is a known contaminant of chrysotile asbestos [Churg et al., 1984], which is the most common type of asbestos used commercially in the United States. Analysis of lung tissue from chrysotile miners and millers with malignant mesothelioma shows prefercntial retention of the contaminating tremolite fibers (Churg et al., 1984). This is probably due to the tendency for chrysotile to break apart within the lungs into individual fibrils, which can then be more readilycleared [Roggli and Brody, 1984). The study by McDonald et al. [1989] indicated that after excluding chrysotile miners and millers, -22% of Canadian mesotheliomas are attributable to tremol'ue. Only 3% of the total fibers in our study were long (> 5 µan) ehrysotile fibers. This value of course does not include the much larger number of short chrysottle fibers, which are frequently found in lung samples from the general population [Churg and Warnock, 1980]. Studies from our laboratory have shown that chrysotik fibers 5 µm or less in length are rapidly cleared from the lungs of rats exposed by inhalation, whereas longer chrysotile fibers are preferentially Peteined [Coin et al., 1992). Kimizuka et al. [1987] reported further fragmentation Qf ioeg thin chrysotile fibers 2 years postexposure in hamsters, with a concomitant in6ease in the percentage of fibers less than 5 µm in length. These short fibers would then be expected to be rapidly cleared from the lungs. Tremolite fibers found in the lung tissues of our mesothelioma cases are probably a marker for the much greater numbers of chryso6le fibers, which were deposited but subsequently cleared. Rogers et al. (1991) concluded that the chrysotile lung burden contributed significantly to the risk of mesothelioma in Australia, whereas McDonald et al. [1989] found no difference in chrysotile content between cases and controls. We are unable to exclude a role for the long chrysotile fibers (> 5 µm) found in our cases in the production of nxsotheliorna, especially since they probably constitute only a minute proportion of the long fibers actually deposited. Nonasbestos mineral fibers accounted for 25% of the total fibers identified in the present study (Table lll). Such fibers are commonly found among members of the general population [Churg, 1983 ; Roggli. 1989a) and account for nearly 80% of fibers 5 µm or greater in length found in lungs of individuals with no identifiable exposure to asbestos (Roggli et al., 1992). Most of these fibers had a low aspect ratio, and very few of them met the Stanton criteria for the most dangerous fibers, i.e., =Y 8 µm in length and < 0.25 µm in diameter. (Fibers that are z 8 µm in length and < 0.25 µm in diameter were the most efficient at producing mesothel'iomas by means of implantation within the pleural cavities of rats (Stanton et al., 1981].) McDonald et al. [1989] could find no evidence that these nonasbestos mineral fibers contribute to the risk of mesothelioma. It has been suggested that it is the fibers that reach the pleura that are the ones responsible for the development of mesothelioma. In this regard, studies by SEbastien et al. (19771 showed that large numbers of short (< 5 µm) chrysotile fibers aocumulate in the pleura, whereas long (> 5 µm) amphibole fibers tend to accumulate within the lung parenchyma. However, a review of the pertinent human and experimental animal literature indicates that fibers less than 5 µm in length are unlikely to

8 612 Roggli et al. be relevant to the pathogenesis of inesothelioma [Lippmann, 1988]. Furthermore, recent studies have shown that, in addition to short chrysotile fibers, some long amosite fibers also reach the pleura (Dodson et al., 1990). Fibers that exceed 5 µm in length and have diameters less than 0.1 µm may be the ones that are most relevant to the development of malignant mesothelioma in man (Uppmann, 1988), although the pathologic changes induced by large numbers of short fibers have yet to be established in human lungs. It could be argued that, since our technique primarily detects fibers that are 0.2 µm or greater in diameter, we could have overlooked substantial numbers of fibers that are 5 µm or greater in length and 0.1 µm or less in diameter. We think that this possibility is unlikely to affect our results or conclusions, because substantial numbers of crocidolite or amosite fibers less than 0.1 µm in diameter should be accompanied by substantial numbers of fibers 0.2 µm or greater in diameter. This opinion is supported by the observations of Churg and Wiggs (1984) and Warnock (1989], since these investigators used transmission electron microscopy and found ratios of amosite to crocidolite among U.S. mesothelioma cases similar to those found In our study. It is more likely that our analysis would underestimate the numbers of chrysotile fibers 5 µm or greater in length, because chrysotile tends to undergo longitudinal splitting with many fibers, as a result, having diantytdrs kss than 0.1 µm [Roggli and Brody, 1984). ' } There is a growing consensus that it is the fibers that accumulate within the lung that are responsible for the development of asbestos-associated diseases, including mesothelioma (Churg, 1988 ; Wagner and Pooley, 1986]. If this is indeed the case, then the results of the present study indicate that, with respechto the occurrence of malignant mesothelioma in the United States, the order of importance of the various asbestos fiber types is Amosite > Tremolite > Chrysotile nr Crocidolite. Our findings are at odds with the assertion that crocidolite asbestos is responsible for most mesotheliomas in the United States. ACKNOWLEDGMENTS The authors gratefully acknowledge the assistance of Dr. Patrick Coin in the preparation of Figure 1, and Nancy Hall for preparation of the manuscript. REFERENCES Albcns AS. Falkwn O, Goedhals L. Vorobiof DA, Van der Mervrc CA (1988) : Malignanl pleunl mesothelioma: A disease unaffected by current therapeutic maneuvers. J Clin Oncol 6 : Chahinian AP. Pajak TF, Holland 1F, Norton L. Ambinder RM. Mandel EM (1982) : Diffuse malignant mesothelioma : Prospective evaluation of 69 patients. Ann Intern Med 96 : Churg A (1983) : Norosbestos pulmonary mineral fibers in rhe general population. Environ Res 31 :1g Churg A (1988) : Chrysotile, Irernolite, and malignant mesorhelioma in man. Chest 93: Churg A, Warnock ML (1980) : Asbestos fibers in the genenl population. Am Rev Respir Dis 122 :

9 Mesothelioma and Asbestos Fiber Type 613 Churg A, Wiggs B (1984) : Fiber size and number in amphibole asbestos-induced mesolholioma. Am 1 Pathol 115 : Churg A- Wiggs B, Depaoli L, Kampe B, Stevens B(1984) : Lung asbestos eontent in chrysotile workers with mesothelioma. Am Rev Respir Dis 130 : Coin PG. Roggli VL, Brody AR (1992): Deposition, eleannce, and transloqdon of ehrysotik asbestos from peripheral and central regions of the rat lung. Envimn Res 58 : Craighead JE, Mossman BT (1982): Pathogenesis of asbesros-assoeiated diseasts. N Engl J Med 306: Dodson RF, Williams MG, Cocn G, Btollo A, Bianchi C(1990) : Asbestos eontau of lung tiuue, lymph nodes, and pleural plaques from fomter shipyard wotkers. Am Rev Resplr Dk 142 :843-g47. Gaudichet A. Janson X. Monchaux 0. Dufour G, S6bastien P. De f.ajartre A. Bignon 1(1988) : Assessment by analytical microscopy of the total lung fibre burden In tnetmlalioma patients tnnehed with four other pathological series. Ann Occup Hyg 32 (Suppl) : HammarSP. Bolen 1W (1988); Pleural neoplasms, Chap. 30. In Dail DH, HammaSP(eds) : "Pulmonary Pathology." New York: Springer-Vedag, pp g : Malignant mesdhelioma 1982 : Review of. Hillerdal G(1983) 4710 published cases. Br J Dis Chest 77 : Kimizuka G, Wang N-S. Hayashi Y(1987) : Physical and microchemkal alterations of ehrysotik and amosite asbestos in the hamster lung. 1 Toxicol Environ Hlth 21 : Lippman M(1988): Asbestos exposure indices. Environ Res 46:86-1(16. McCaughey WTE, Colby TV, Battifon H. Churg A. C.oson JM. Greenberg SD, Grimes MM. Hammar S, Roggli VL, Unni KK (1991) : The diagnosis of diffuse malignant IrKsotheli6rna: Experience or the U.S./Canadian Mesothelioma Panel. Mod Patlol 4 : Armstrong B, Case B. Doell D. McCaughey WTE, McDonald. JC AD ; Sdbaslien P(1989): Mesothelioma and asbestos fiber type : Evidence from lung tissue analysei'gncer 63 : Mossman BT, Bignon 1, Corn M, Seaton A, Cxe 1BL (1990) : Asbestos: Scientific develoymenls and implications for public policy. Science 247 : Nicholson Wl. Johnson EM. Harington JS, Melius J, Landrigan PJ (1990) : Asbestos, carcinogenicity and public policy (kner). Science 248 : Rogers Al. Leigh J, Bcrry G, Ferguson DA, Mulder HB, Ackad M(1991) : Relationship between lung asbestos fiber type and concentration and relative risk of inesotmlioma : A ease-eomrol study. Cancer 67 : : Nonasbestos. Roggli VL (I989a) mineral fibers in human lungs. In : Russell PE (ed) : "Mierobeam Analysis-1989." San Francisco : San Francisco Press, pp Roggli VL (1989b) : Scanning electron microscopic analysis of mineral fibers in human Wngs. Chap. 5. In Ingram P. Shelburne JD, Roggli VL (eds) : "Microprobe Analysis in Medieine." New York : Hemisphere, pp Roggli VL (1991) : Mineral fiber content of lung tissue in patients with malignant mesothelioma. Chap. 6. In : Henderson DW, Shilkin KB. Langlois SLP. Whitaker D(eds): "Malignant Mesotheliorna." New York : Hemisphere, pp Roggli VL, Brody AR (1984): Changes in numbers and dimensions of chrysotile asbestos fibers in lungs or rxls following shon-term exposure. Exp Lung Res 7 : Roggli VL. Kolbeck 1, Sanfilippo F. Shelburne 1D (1987); Pathology of human mesothelioma: Etiologic and diagnostic considerations. Pathol Annu 22 (2): Roggli VL, McGavran MH, Subach JA, Sybers HD, Greenberg SD (1982): Pulmonary asbestos body counts and electron probe analysis of asbestos body cores in patients with mesothelioma : A study of 25 cases. Cancer 50 : )2. Roggli VL, Pratt PC, Brody AR (1986) : Asbcstos content of lung tissue in asbestos-associated diseases : A study of 110 cases. Br I Ind Med 43 : Roggli VL, Pran PC. Brody AR (1992) : Analysis of tissue mineral fiber content, Chap 11. In Roggli VL. Greenberg SD, Pratt PC (eds) : "Pathology of Asbestos-ASSociated Diseases." Boston : Little, Brown, pp Ruffic P. Feld R, Minkin S, Cormier Y, Boutan-Larou A, Ginsberg R, Ayoub J, Shepherd FA, Evans WK, Figueredo A. Pater JL. Pringlc 1F. Krcisman H (1989): Diffuse malignant mesothelioma of the pleura in Ontario and Quebec : A retrospective study of 332 patients. J Clin Oncol 7: SEbastien P. Fondimare A, Bignon 1, Monchaux G, Desbordes 1, Bonnaud G(1977) : Topographic

10 614 Roggli et al. _ distribution of asbestos fibers In human lung in rcfation to occupational and ranoccupational exposure. In Walton WH. McGovern B(eds): ' Inhaled Partieks IV." Oxford : Pergamon Presa, pp ' Stanton MF, layard M. Tegeris A. Miller E. May M, Morgan E, Smith A(I9g1) : Relation of qrtkie dimensions to carcinogenicity in amphibole asbestotes and dhee fibnus minerals. J Natl Cancer Inst 67 : Wagner JC (1991a) : Foreword. In Henderson DW, Shilkin KB. LsngkSs SLP, Whitaka D(eds): "Ma lignant Mesahelioma." New Yosk: Hemisphese, pp xvii-uv. Wagner JC (1991b) : The discovcry of the association between blue asbestos and mestnheliumas and Ihe aftermath. Br I Ind Med 48 : Wagner JC, Pooky FU (1986) : Mineral fibres and mesoslaiioma. 71qnx 41 : Wagner JC, Berry G, Skidmore 1 W, Timbrell V(197Q: The t(fecu or the inhalation of asbestos in nts. Br I Cancer 29 : Warnock ML (1989) : Lung asbestos burden in shipyard and construction workess with snesowelioma : Comparison with burdens in subjects with asbestosis or lung cancer. Environ Res 50:68-g5.